Hi Paul,
This one (InsiderAdvantage) wasn’t too far off: Obama 34, Edwards 33, Clinton 32; Huckabee 30, Romney 24, Thompson/McCain 11.
This one (Reuters-CSPAN-Zogby) was also fairly close, although low on all counts: Obama 31, Edwards 27, Clinton 24; Huckabee 31, Romney 25, Thompson 11, McCain 10.
This one (American Research Group) was close for the GOP but off for the Dems bigtime (Clinton 34, Obama 25, Edwards 21).
- Jeremy

I recognize the difficulty of these kinds of polls. Nevertheless I imagine that some pollsters are more sensitive to the constantly evolving factors that reflect the actual likely behavior of the participants such as young folks who don’t have landline phones or independents who don’t follow party lines.

Apparently independents had a big impact on Obama’s results in Iowa.

I don’t think there’s a magic bullet. I do think RCP’s aggregated, state-by-state breakdowns are the best place to start, as you do, but then there’s a big “zeitgeist” factor that plays into this, obviously, and that’s where I rely TMV, Politico, etc.

What I would NOT rely on is traditional assumptions about campaign strategy. My boss referenced in this post actually scored worse than any of us in his little survey, largely (I’m convinced) because he assumed the best on-the-ground staff/organization would keep Romney and Clinton at the top of their respective ballots. That’s a decent assumption, but it completely fails to appreciate the power in this election of what I like to call the “change quotient” and the “genuine quotient,” neither of which Mitt or Hillary could offer, as was clear from posts in these three places prior to the actual caucuses.

My two cents.